For some peculiar reason I'm involved with three customers at the same time wanting me to assist them in changing their way of doing systems development. And I say peculiar, because although this is my expertise for many years already, there has not been even one customer asking me to help them with this for almost five years. Maybe it's the sign of the time, or maybe it is because I stopped dying my hair letting my gray hairs come through and customers finally start to take me seriously. I don't know.
Anyway. It is at the same time I see many people turning away from a Big Design Up Front, plan-based methodologies or (at the other side of the spectrum) from working without any formal method at all. Both have in common that they have an urgent need to get a better grip on the (planning of) the development process, and it is likely they will meet each other somewhere 'in the middle' where the agile methodologies are.
It's funny to notice that getting started with agile methodologies in itself also is best done in an agile way. With that I mean starting with a minimal set and from there, discover where and when you need to scale up. But although you start with a minimal set it still is important you are aware of which direction you want to go. It is even more important to ensure that the people that will be doing the actual work, agree this is the direction you all want to go. Otherwise the risk of resistance up to a level where you will fail misserably will be unacceptable high.
For that reason many of the techniques I suggest to introduce into the development proces, I actually use for introducing the techniques themselves. Like a meta-method if you will. Depending on what I think would fit, this could be using the MoSCoW principle for prioritization the changes. But also make them aware of the similarities between the Onsite Customer practice and them being involved in changing the development process.
This has many advantages. The most important ones are that I verify early in the process if people actually understand the technique and if it will work for them. While doing so I have the opportunity to use a kind of trial and error way of implementing a technique, and securing it not only in a formal way but also by creating enough support for it to last after I have left the building.